Chris,
I have finally got back to my mac to test this. I have installed
the latest j806 to make sure it's up to date (and tried this with j805
with the same results as reported here). Both pass your test:
F=: jpath '~temp/t1.sh'
F fwrites~ '#!/bin/bash',LF,'sleep 2',LF,'echo "done"'
32
fmakex F
foo=: 3 : 0
echo 2!:0 F
echo 2+3
)
foo''
done
5
The mac is a late 2014 mini running OS 10.12.6 (Sierra).
JVERSION
Engine: j806/j64nonavx/darwin
Release: commercial/2017-11-06T10:18:00
Library: 8.06.09
Qt IDE: 1.6.2s/5.6.3
Platform: Darwin 64
Installer: J806 install
InstallPath: /users/jph/j64-806
But... my program still fails. I can simplify the problem by just entering
one line (after setting the directory to /Users/jph/tlusty205/synspec/jph):
2!:0 '../synmod.exe < fort.5 > fort.6'
Instead of writing the output to fort.6, it dumps that output to the Jqt
session:
...
PARTITION FUNCTIONS AT THE STANDARD DEPTH
------------------------------------------
H 4.58E+00 1.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
0.00E+00
He 1.00E+00 2.00E+00 1.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
0.00E+00
Li 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
0.00E+00
Be 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
0.00E+00
B 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
0.00E+00
C 1.19E+02 6.12E+00 1.04E+00 2.01E+00 1.00E+00 0.00E+00 0.00E+00
0.00E+00
etc.
This is just the tail of the 557 line output file. So the > direction of
the script is not followed. But if I go to a terminal shell in the same
directory and enter
../synmod.exe < fort.5 > fort.6
it executes without any screen output. And the results are printed to
fort.6.
On my Dell XPS running 806 under Ubuntu 16.04,
2!:0 '../synmod.exe < fort.5 > fort.6'
returns to the qt terminal without any screen output, and looking at
fort.6, the expected results have been written there. So there is
some problem about how 2!:0 handles the directed output on the mac as
opposed to the Linux machine.
I suppose I could work around this by modifing the fortran program to send
the output directly to a file called 'fort.6' instead of directing the
standard output. But it's huge program I didn't write, and I'd rather not.
Regards, Patrick
On Wed, 22 Nov 2017, chris burke wrote:
You should be able to do a simple test to see if 2!:0 waits, for example as
below. This does wait on my Mac with the same J as you, except the avx
version.
F=: jpath '~temp/t1.sh'
F fwrites~ '#!/bin/bash',LF,'sleep 2',LF,'echo "done"'
fmakex F
foo=: 3 : 0
echo 2!:0 F
echo 2+3
)
foo''
On Wed, Nov 22, 2017 at 7:28 PM, J. Patrick Harrington <[email protected]>
wrote:
Bill,
I've tried that. ../synmod.exe < fort.5 > fort.6 in a terinal
window works fine on all three machines. And the directory change
is OK. WHen the program fails on the Mac Mini, the J session is
left in the /home/tages/tlusty205/synspec/jph directory.
Patrick
On Thu, 23 Nov 2017, bill lam wrote:
You can try executing the cmd in bash shell to verify whether it actually
work on the other machine.
Also check the cwd in J session to see it is correct.
On Nov 23, 2017 12:54 AM, "J. Patrick Harrington" <[email protected]>
wrote:
A>> A problem has surfaced which has me puzzled. I have written a
program which needs the results of another program, synmod.exe, which
reads in a file, executes for a few seconds, and writes the results
to the file fort.6 (synmod.exe is my modification of a 10,000 line
fortran program). The relevant part of my program is as follows:
LAM=: IX=: QX=: <''
1!:44 '/home/tages/tlusty205/synspec/jph'
'Ti gi lam1 lam2'=. y
fix55 lam1, lam2 NB. put lam1 & lam2 into fort.55
lngg=. 10^. gi
nh=. -:#Ti
i=. _1
while. (i=. >:i)<nh do.
Tii=. i{Ti
lngi=. i{lngg
Kbicub Tii, lngi
2!:0 '../synmod.exe < fort.5 > fort.6'
Z=. lam2 CLEAN Extract''
'mu rad deg lam Flx Ia Qa pol'=. Z
LAM=: LAM, <lam
IX=: IX, <Ia
QX=: QX, <Qa
end.
1!:44 '/home/tages/J6/SPIN.d'
The issue here is the Foreign 2!:0. This code works *perfectly* on my
Dell
laptop running
JVERSION
Engine: j806/j64/linux
Release: commercial/2017-11-06T09:54:01
Library: 8.06.09
Qt IDE: 1.6.2/5.3.2
Platform: Linux 64
Installer: J806 install
InstallPath: /home/tages/j64-806
But... the very same code running on either of the other two machines I
use fails. The line 2!:0 '../synmod.exe < fort.5 > fort.6'
should write the results of synmod to the file fort.6, but
instead defines fort.6 but writes nothing to it. Thus the
Extract'' routine fails when it tries to read fort.6.
I suspect that while 2!:0 waits for synmod.exe to finish, maybe
on my other machines it transfers to the next line of the J code
before fort.6 is written. One of these machines is a Mac Mini running
JVERSION
Engine: j806/j64nonavx/darwin
Release: commercial/2017-11-06T10:18:00
Library: 8.06.09
Qt IDE: 1.6.2s/5.6.3
Platform: Darwin 64
Installer: J806 install
InstallPath: /users/jph/j64-806
So I'm asking if the foreign 2!:0 in some cases returns without
waiting. I thought just 2!:1 did that.
Apologies for the long-winded post, Patrick
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
